Expert system for materials design of PTC thermistors

This expert system has been developed for helping engineers to design the PTC (Positive Temperature Coefficient) thermistors which are typical electrical ceramics. The engineer presents the composition of a PTC thermistor, and this system predicts the properties, such as switching temperature, resistivity at room temperature, withstanding voltage and density. This system consists of 4 blocks. (1)Composition-input block; Users are required to input the compositions of main constituent elements and typical additives. (2)First(qualitative)-inference block; This block shows the simple inferences, such as “non-conductive”, “high density”. This block is represented by using IF-THEN rules, so the knowledge can be added, removed, and revised very easily. (3)Database-search block. (4)Second(quantitative)-inference block; This block shows the numerical values of the properties. It uses the empirical equations where coefficients are optimized from data by simplex method.